README.md

Thermolysin

This folder contains instructions and scripts for analyzing anomalous signal from an XFEL dataset of thermolysin (CXIDB, entry 81; Dalton et al. 2022, Nat. Comm. 2022, https://doi.org/10.1038/s41467-022-35280-8). Here, we

1. run careless with a bivariate prior, for many values of the double-Wilson r parameter, and then
2. inspect merging statistics and anomalous difference peak heights.

We start with MTZ files found in ./unmerged_mtzs. These were converted from CXIDB entry 81.

Configuring the bivariate prior

To run careless, we use the script careless_runs/slurm-dw-array-grid.sh, which starts a slurm batch array job. This job requires careless_runs/slurm_params.txt, in which we vary the double-Wilson r value across the individual careless runs. To call using slurm:

cd careless_runs
sbatch slurm-dw-array-grid.sh

Two flags in slurm-dw-array-grid.sh control the behavior of the bivariate prior:

CARELESS_ARGS+=(--double-wilson-parents=${DW_LIST}) 
CARELESS_ARGS+=(--double-wilson-r=${DWR_LIST})

DW_LIST here is None,0. This constructs our Bayesian network. The index for each entry refers to the corresponding input MTZ. The value of each entry corresponds to the node that is the ``parent'' of that entry. In this case, the 0th node does not have a parent, while the parent of the 1st node is node 0. The unmerged input MTZs are the positive and negative Friedel mates. Graphically, their dependency looks like this:

while DWR_LIST=0.,0.99, for example, lists the values of the correlation parameter $r_{DW}$ associated with each edge, so

Since the first edge points to None, we set its correlation parameter arbitrarily to 0.

Further notes

Many bash scripts require activating a conda environment with careless in it. Please take note that you are activating the right conda environment! Additionally, the two refinement scripts, scripts/launch_refinement.sh and scripts/launch_refinement_omit.sh, both require sourcing your copy of phenix_env.sh, the environment script that comes with your copy of phenix.

After running careless, we evaluate the quality of the careless results in the jupyter notebook titled Inspect_Careless_param_grid.ipynb. This notebook also plots figures.

Folders

• unmerged_mtzs: a folder with three unmerged MTZ files. unmerged.mtz contains unmerged intensities from CXIDB entry 81. friedel_{plus,minus}.mtz are the outputs of scripts/friedelize.py, which splits unmerged.mtz into F+ and F- half-datasets.
• careless_runs: a folder containing a script for running careless as a batch array, as well as the resultant subfolders containing outputs from individual runs of careless.
• refinement: a folder containing reference pdb files as well as phenix .eff files for refinement of careless outputs. Additionally, the folder contains an MTZ file with reference R-free flags.
• scripts: a folder containing scripts that are used for processing the output data. Included in scripts are:
  • anomalous_peak_heights.py: a script called by run_ccs.sh that computes the anomalous peak heights at the five anomalous scattering positions.
  • friedelize.py: the script used for splitting unmerged.mtz into F+ and F- half-datasets.
  • launch_refinement_omit.sh: a script for refining an omit model against careless outputs, where the model's anomalous scatterers have occupancy set to 0. This script is called in unfriedelize_all.sh.
  • launch_refinement.sh: a script for refining a model against against careless outputs. This script is called in unfriedelize_all.sh.
  • run_ccs.sh: a script for calculating CC$\text{1/2}$, CC${\text{pred}}$, CC$_\text{anom}$, and anomalous peak heights for each careless output. This script is called in Inspect_Careless_param_grid.ipynb.
  • unfriedelize_all.sh: A script that converts output careless MTZ files into downstream readable files. Additionally, this script starts refinement of the careless outputs. This script is called in Inspect_Careless_param_grid.ipynb.
  • unfriedelize.py: a script for unsplitting MTZ files that have been split by friedelize.py.